includes/filerepo/backend/FileBackendMultiWrite.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup FileBackend
   5  * @author Aaron Schulz
   6  */
   7
   8 /**
   9  * @brief Proxy backend that mirrors writes to several internal backends.
  10  *
  11  * This class defines a multi-write backend. Multiple backends can be
  12  * registered to this proxy backend and it will act as a single backend.
  13  * Use this when all access to those backends is through this proxy backend.
  14  * At least one of the backends must be declared the "master" backend.
  15  *
  16  * Only use this class when transitioning from one storage system to another.
  17  *
  18  * Read operations are only done on the 'master' backend for consistency.
  19  * Write operations are performed on all backends, in the order defined.
  20  * If an operation fails on one backend it will be rolled back from the others.
  21  *
  22  * @ingroup FileBackend
  23  * @since 1.19
  24  */
  25 class FileBackendMultiWrite extends FileBackend {
  26         /** @var Array Prioritized list of FileBackendStore objects */
  27         protected $backends = array(); // array of (backend index => backends)
  28         protected $masterIndex = -1; // integer; index of master backend
  29         protected $syncChecks = 0; // integer bitfield
  30
  31         /* Possible internal backend consistency checks */
  32         const CHECK_SIZE = 1;
  33         const CHECK_TIME = 2;
  34
  35         /**
  36          * Construct a proxy backend that consists of several internal backends.
  37          * Additional $config params include:
  38          *     'backends'    : Array of backend config and multi-backend settings.
  39          *                     Each value is the config used in the constructor of a
  40          *                     FileBackendStore class, but with these additional settings:
  41          *                         'class'         : The name of the backend class
  42          *                         'isMultiMaster' : This must be set for one backend.
  43          *     'syncChecks'  : Integer bitfield of internal backend sync checks to perform.
  44          *                     Possible bits include self::CHECK_SIZE and self::CHECK_TIME.
  45          *                     The checks are done before allowing any file operations.
  46          * @param $config Array
  47          */
  48         public function __construct( array $config ) {
  49                 parent::__construct( $config );
  50                 $namesUsed = array();
  51                 // Construct backends here rather than via registration
  52                 // to keep these backends hidden from outside the proxy.
  53                 foreach ( $config['backends'] as $index => $config ) {
  54                         $name = $config['name'];
  55                         if ( isset( $namesUsed[$name] ) ) { // don't break FileOp predicates
  56                                 throw new MWException( "Two or more backends defined with the name $name." );
  57                         }
  58                         $namesUsed[$name] = 1;
  59                         if ( !isset( $config['class'] ) ) {
  60                                 throw new MWException( 'No class given for a backend config.' );
  61                         }
  62                         $class = $config['class'];
  63                         $this->backends[$index] = new $class( $config );
  64                         if ( !empty( $config['isMultiMaster'] ) ) {
  65                                 if ( $this->masterIndex >= 0 ) {
  66                                         throw new MWException( 'More than one master backend defined.' );
  67                                 }
  68                                 $this->masterIndex = $index;
  69                         }
  70                 }
  71                 if ( $this->masterIndex < 0 ) { // need backends and must have a master
  72                         throw new MWException( 'No master backend defined.' );
  73                 }
  74                 $this->syncChecks = isset( $config['syncChecks'] )
  75                         ? $config['syncChecks']
  76                         : self::CHECK_SIZE;
  77         }
  78
  79         /**
  80          * @see FileBackend::doOperationsInternal()
  81          * @return Status
  82          */
  83         final protected function doOperationsInternal( array $ops, array $opts ) {
  84                 $status = Status::newGood();
  85
  86                 $performOps = array(); // list of FileOp objects
  87                 $paths = array(); // storage paths read from or written to
  88                 // Build up a list of FileOps. The list will have all the ops
  89                 // for one backend, then all the ops for the next, and so on.
  90                 // These batches of ops are all part of a continuous array.
  91                 // Also build up a list of files read/changed...
  92                 foreach ( $this->backends as $index => $backend ) {
  93                         $backendOps = $this->substOpBatchPaths( $ops, $backend );
  94                         // Add on the operation batch for this backend
  95                         $performOps = array_merge( $performOps, $backend->getOperations( $backendOps ) );
  96                         if ( $index == 0 ) { // first batch
  97                                 // Get the files used for these operations. Each backend has a batch of
  98                                 // the same operations, so we only need to get them from the first batch.
  99                                 $paths = $backend->getPathsToLockForOpsInternal( $performOps );
 100                                 // Get the paths under the proxy backend's name
 101                                 $paths['sh'] = $this->unsubstPaths( $paths['sh'] );
 102                                 $paths['ex'] = $this->unsubstPaths( $paths['ex'] );
 103                         }
 104                 }
 105
 106                 // Try to lock those files for the scope of this function...
 107                 if ( empty( $opts['nonLocking'] ) ) {
 108                         // Try to lock those files for the scope of this function...
 109                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 110                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 111                         if ( !$status->isOK() ) {
 112                                 return $status; // abort
 113                         }
 114                 }
 115
 116                 // Clear any cache entries (after locks acquired)
 117                 $this->clearCache();
 118
 119                 // Do a consistency check to see if the backends agree
 120                 if ( count( $this->backends ) > 1 ) {
 121                         $status->merge( $this->consistencyCheck( array_merge( $paths['sh'], $paths['ex'] ) ) );
 122                         if ( !$status->isOK() ) {
 123                                 return $status; // abort
 124                         }
 125                 }
 126
 127                 // Actually attempt the operation batch...
 128                 $subStatus = FileOp::attemptBatch( $performOps, $opts, $this->fileJournal );
 129
 130                 $success = array();
 131                 $failCount = 0;
 132                 $successCount = 0;
 133                 // Make 'success', 'successCount', and 'failCount' fields reflect
 134                 // the overall operation, rather than all the batches for each backend.
 135                 // Do this by only using success values from the master backend's batch.
 136                 $batchStart = $this->masterIndex * count( $ops );
 137                 $batchEnd = $batchStart + count( $ops ) - 1;
 138                 for ( $i = $batchStart; $i <= $batchEnd; $i++ ) {
 139                         if ( !isset( $subStatus->success[$i] ) ) {
 140                                 break; // failed out before trying this op
 141                         } elseif ( $subStatus->success[$i] ) {
 142                                 ++$successCount;
 143                         } else {
 144                                 ++$failCount;
 145                         }
 146                         $success[] = $subStatus->success[$i];
 147                 }
 148                 $subStatus->success = $success;
 149                 $subStatus->successCount = $successCount;
 150                 $subStatus->failCount = $failCount;
 151
 152                 // Merge errors into status fields
 153                 $status->merge( $subStatus );
 154                 $status->success = $subStatus->success; // not done in merge()
 155
 156                 return $status;
 157         }
 158
 159         /**
 160          * Check that a set of files are consistent across all internal backends
 161          *
 162          * @param $paths Array
 163          * @return Status
 164          */
 165         public function consistencyCheck( array $paths ) {
 166                 $status = Status::newGood();
 167                 if ( $this->syncChecks == 0 ) {
 168                         return $status; // skip checks
 169                 }
 170
 171                 $mBackend = $this->backends[$this->masterIndex];
 172                 foreach ( array_unique( $paths ) as $path ) {
 173                         $params = array( 'src' => $path, 'latest' => true );
 174                         // Stat the file on the 'master' backend
 175                         $mStat = $mBackend->getFileStat( $this->substOpPaths( $params, $mBackend ) );
 176                         // Check of all clone backends agree with the master...
 177                         foreach ( $this->backends as $index => $cBackend ) {
 178                                 if ( $index === $this->masterIndex ) {
 179                                         continue; // master
 180                                 }
 181                                 $cStat = $cBackend->getFileStat( $this->substOpPaths( $params, $cBackend ) );
 182                                 if ( $mStat ) { // file is in master
 183                                         if ( !$cStat ) { // file should exist
 184                                                 $status->fatal( 'backend-fail-synced', $path );
 185                                                 continue;
 186                                         }
 187                                         if ( $this->syncChecks & self::CHECK_SIZE ) {
 188                                                 if ( $cStat['size'] != $mStat['size'] ) { // wrong size
 189                                                         $status->fatal( 'backend-fail-synced', $path );
 190                                                         continue;
 191                                                 }
 192                                         }
 193                                         if ( $this->syncChecks & self::CHECK_TIME ) {
 194                                                 $mTs = wfTimestamp( TS_UNIX, $mStat['mtime'] );
 195                                                 $cTs = wfTimestamp( TS_UNIX, $cStat['mtime'] );
 196                                                 if ( abs( $mTs - $cTs ) > 30 ) { // outdated file somewhere
 197                                                         $status->fatal( 'backend-fail-synced', $path );
 198                                                         continue;
 199                                                 }
 200                                         }
 201                                 } else { // file is not in master
 202                                         if ( $cStat ) { // file should not exist
 203                                                 $status->fatal( 'backend-fail-synced', $path );
 204                                         }
 205                                 }
 206                         }
 207                 }
 208
 209                 return $status;
 210         }
 211
 212         /**
 213          * Substitute the backend name in storage path parameters
 214          * for a set of operations with that of a given internal backend.
 215          *
 216          * @param $ops Array List of file operation arrays
 217          * @param $backend FileBackendStore
 218          * @return Array
 219          */
 220         protected function substOpBatchPaths( array $ops, FileBackendStore $backend ) {
 221                 $newOps = array(); // operations
 222                 foreach ( $ops as $op ) {
 223                         $newOp = $op; // operation
 224                         foreach ( array( 'src', 'srcs', 'dst', 'dir' ) as $par ) {
 225                                 if ( isset( $newOp[$par] ) ) { // string or array
 226                                         $newOp[$par] = $this->substPaths( $newOp[$par], $backend );
 227                                 }
 228                         }
 229                         $newOps[] = $newOp;
 230                 }
 231                 return $newOps;
 232         }
 233
 234         /**
 235          * Same as substOpBatchPaths() but for a single operation
 236          *
 237          * @param $op File operation array
 238          * @param $backend FileBackendStore
 239          * @return Array
 240          */
 241         protected function substOpPaths( array $ops, FileBackendStore $backend ) {
 242                 $newOps = $this->substOpBatchPaths( array( $ops ), $backend );
 243                 return $newOps[0];
 244         }
 245
 246         /**
 247          * Substitute the backend of storage paths with an internal backend's name
 248          *
 249          * @param $paths Array|string List of paths or single string path
 250          * @param $backend FileBackendStore
 251          * @return Array|string
 252          */
 253         protected function substPaths( $paths, FileBackendStore $backend ) {
 254                 return preg_replace(
 255                         '!^mwstore://' . preg_quote( $this->name ) . '/!',
 256                         StringUtils::escapeRegexReplacement( "mwstore://{$backend->getName()}/" ),
 257                         $paths // string or array
 258                 );
 259         }
 260
 261         /**
 262          * Substitute the backend of internal storage paths with the proxy backend's name
 263          *
 264          * @param $paths Array|string List of paths or single string path
 265          * @return Array|string
 266          */
 267         protected function unsubstPaths( $paths ) {
 268                 return preg_replace(
 269                         '!^mwstore://([^/]+)!',
 270                         StringUtils::escapeRegexReplacement( "mwstore://{$this->name}" ),
 271                         $paths // string or array
 272                 );
 273         }
 274
 275         /**
 276          * @see FileBackend::doPrepare()
 277          * @return Status
 278          */
 279         protected function doPrepare( array $params ) {
 280                 $status = Status::newGood();
 281                 foreach ( $this->backends as $backend ) {
 282                         $realParams = $this->substOpPaths( $params, $backend );
 283                         $status->merge( $backend->doPrepare( $realParams ) );
 284                 }
 285                 return $status;
 286         }
 287
 288         /**
 289          * @see FileBackend::doSecure()
 290          * @return Status
 291          */
 292         protected function doSecure( array $params ) {
 293                 $status = Status::newGood();
 294                 foreach ( $this->backends as $backend ) {
 295                         $realParams = $this->substOpPaths( $params, $backend );
 296                         $status->merge( $backend->doSecure( $realParams ) );
 297                 }
 298                 return $status;
 299         }
 300
 301         /**
 302          * @see FileBackend::doClean()
 303          * @return Status
 304          */
 305         protected function doClean( array $params ) {
 306                 $status = Status::newGood();
 307                 foreach ( $this->backends as $backend ) {
 308                         $realParams = $this->substOpPaths( $params, $backend );
 309                         $status->merge( $backend->doClean( $realParams ) );
 310                 }
 311                 return $status;
 312         }
 313
 314         /**
 315          * @see FileBackend::getFileList()
 316          */
 317         public function concatenate( array $params ) {
 318                 // We are writing to an FS file, so we don't need to do this per-backend
 319                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 320                 return $this->backends[$this->masterIndex]->concatenate( $realParams );
 321         }
 322
 323         /**
 324          * @see FileBackend::fileExists()
 325          */
 326         public function fileExists( array $params ) {
 327                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 328                 return $this->backends[$this->masterIndex]->fileExists( $realParams );
 329         }
 330
 331         /**
 332          * @see FileBackend::getFileTimestamp()
 333          */
 334         public function getFileTimestamp( array $params ) {
 335                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 336                 return $this->backends[$this->masterIndex]->getFileTimestamp( $realParams );
 337         }
 338
 339         /**
 340          * @see FileBackend::getFileSize()
 341          */
 342         public function getFileSize( array $params ) {
 343                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 344                 return $this->backends[$this->masterIndex]->getFileSize( $realParams );
 345         }
 346
 347         /**
 348          * @see FileBackend::getFileStat()
 349          */
 350         public function getFileStat( array $params ) {
 351                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 352                 return $this->backends[$this->masterIndex]->getFileStat( $realParams );
 353         }
 354
 355         /**
 356          * @see FileBackend::getFileContents()
 357          */
 358         public function getFileContents( array $params ) {
 359                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 360                 return $this->backends[$this->masterIndex]->getFileContents( $realParams );
 361         }
 362
 363         /**
 364          * @see FileBackend::getFileSha1Base36()
 365          */
 366         public function getFileSha1Base36( array $params ) {
 367                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 368                 return $this->backends[$this->masterIndex]->getFileSha1Base36( $realParams );
 369         }
 370
 371         /**
 372          * @see FileBackend::getFileProps()
 373          */
 374         public function getFileProps( array $params ) {
 375                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 376                 return $this->backends[$this->masterIndex]->getFileProps( $realParams );
 377         }
 378
 379         /**
 380          * @see FileBackend::streamFile()
 381          */
 382         public function streamFile( array $params ) {
 383                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 384                 return $this->backends[$this->masterIndex]->streamFile( $realParams );
 385         }
 386
 387         /**
 388          * @see FileBackend::getLocalReference()
 389          */
 390         public function getLocalReference( array $params ) {
 391                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 392                 return $this->backends[$this->masterIndex]->getLocalReference( $realParams );
 393         }
 394
 395         /**
 396          * @see FileBackend::getLocalCopy()
 397          */
 398         public function getLocalCopy( array $params ) {
 399                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 400                 return $this->backends[$this->masterIndex]->getLocalCopy( $realParams );
 401         }
 402
 403         /**
 404          * @see FileBackend::getFileList()
 405          */
 406         public function getFileList( array $params ) {
 407                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 408                 return $this->backends[$this->masterIndex]->getFileList( $realParams );
 409         }
 410
 411         /**
 412          * @see FileBackend::clearCache()
 413          */
 414         public function clearCache( array $paths = null ) {
 415                 foreach ( $this->backends as $backend ) {
 416                         $realPaths = is_array( $paths ) ? $this->substPaths( $paths, $backend ) : null;
 417                         $backend->clearCache( $realPaths );
 418                 }
 419         }
 420 }